ReadMe File for “Confronting Core Issues: A Critical Assessment of Attitude Polarization using Tailored Experiments” 

Replication.Rmd reproduces all of the tables and figures provided in the main text and appendix using the statistical programming language, R. “Replication.html” produces output (e.g., figures, statistical output) for all of the scripts featured in the .Rmd file. Scripts to perform the main and appendix analyses are directly provided in the Rmd file.

Likewise, Replication_ASD.Rmd reproduces all of the tables and figures provided in the "Additional Study Details" supplementary text. “Replication_ASD.html” produces output for all scripts included in the .Rmd. 

Both Replication.Rmd and Replication_ASD.Rmd make calls to the following data cleaning scripts in the “scripts” folder: clean_study1.R, clean_study2.R, clean_study3.R, clean_study4.R, and clean_study5.R. Additional routines to prepare the appendix analyses can be found in scripts/prep_appendix_B1.R, scripts/prep_appendix_B2.R, and scripts/prep_appendix_B3.R.

R Version: 4.2.3

Required R packages: 
tidyverse 2.0.0
tidytext 0.4.1
MASS 7.3-60.0.1
moments 0.14.1
ggcorrplot 0.1.4.1
mirt 1.41
lattice 0.22-6
psych 2.4.3
stargazer 5.2.3
modelsummary 2.0.0
texreg 1.39.3
broom 1.0.5
hrbrthemes 0.8.7
vader 0.2.1
labelled 2.12.0
expss 0.11.6
maditr 0.8.4
stm 1.3.7
estimatr 1.0.4
emmeans 1.10.1
magrittr 2.0.3
qualtRics 3.2.0

Raw source files
===========
data/cloudresearch_demographics.csv: Racial and ethnic demographic estimates for Experiment 1 and 2, obtained by recontacting a random sample of participants from each study (N = 517)
data/cloudresearch_kialo_ratings.csv: Qualtrics survey data from the Kialo argument strength study (see Appendix B.2)
data/consistency_handcoded.csv: Handcodings for consistency of GPT output (see Replication_ASD.Rmd code for ASD B.6, Tables B5 and B6)
data/kialo_cloudresearch_pairs.csv: Full list of Kialo-GPT argument pairs used for the Kialo argument strength study (see Appendix B.2)
data/measurement_study.csv: Qualtrics survey data from the measurement study (see Appendix B.4)
data/mturk_codes.csv: Mechanical Turk counts of denigrating comments in Experiment 1, collected for validation tests (see Appendix A.1)
data/mturk_codes2.csv: Mechanical Turk counts of denigrating comments in Experiment 2, collected for analysis of disconfirmation bias (see Appendix A.4)
data/study1.csv: Qualtrics survey data from Experiment 1
data/study1_handcoded.csv: CAP subtopic codings for Experiment 1, generated using GPT-3 finetuned with Experiment 2 subtopics
data/study2_w1.csv: Qualtrics survey data from Experiment 2, Wave 1
data/study2_w2.csv: Qualtrics survey data from Experiment 2, Wave 2
data/study3.csv: Qualtrics survey data from Experiment 3
data/study3_aligned_handcoded.csv: Handcodings for "true_counter" for Experiment 3 (see Replication_ASD.Rmd code for ASD B.6, Table B7)
data/study3_handcoded.csv: CAP subtopic codings for Experiment 3, generated using GPT-3 finetuned with Experiment 2 subtopics
data/study4.csv: Qualtrics survey data from Experiment 4
data/study5.csv: Qualtrics survey data from Experiment 5